Method for predicting viscosity of metallurgical slag through machine learning

Abstract

The invention discloses a method for predicting metallurgical slag viscosity through machine learning, and belongs to the technical field of metallurgical slag viscosity prediction. The method comprises the following steps: acquiring microstructure information of metallurgical slag in a mode of combining experimental detection and molecular dynamics simulation, adding a microstructure as a characteristic variable by adopting a machine learning method on the basis of components and temperature, and establishing a viscosity prediction model of the slag so as to obtain a predicted value of the viscosity of the slag to be detected. According to the method, the microstructure information is obtained by collecting experimental data and performing molecular dynamics simulation, and the molten slag viscosity prediction model based on the microstructure is established by adopting a machine learning method, so that the viscosity of the metallurgical molten slag under different components and different temperature conditions can be quickly and accurately obtained; the precision and the stability of predicting the viscosity of the metallurgical slag are obviously improved; and the cost is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of metallurgical slag viscosity prediction, and in particular to a method for predicting metallurgical slag viscosity through machine learning. Background Art

[0002] The iron and steel industry is an important basic industry on which social and economic development depends, and metallurgical slag plays a vital role in the modern iron and steel smelting process. From blast furnace and converter refining to continuous casting, slag has an important impact on metallurgical quality, smelting efficiency and smooth metallurgical operation. Making full use of the characteristics of slag and rationally regulating its composition and performance can optimize the smelting process, improve metal quality and resource utilization, and promote the development of iron and steel smelting in the direction of high quality, high efficiency, green and safety. Viscosity is one of the physical properties that directly affects the metallurgical function of slag. It characterizes the resistance caused by the internal friction generated by the relative motion between molecules when the slag flows, and is usually used to measure its fluidity. The viscosity of metallurgical slag will directly affect the chemical reaction rate at the slag-metal interface, the forming and surface quality of the ingot, the alloy structure and properties, and the formation of pores in the metal. Appropriate slag viscosity is conducive to the heat transfer process inside the molten pool, thereby promoting the uniform distribution of temperature in the molten pool, reducing energy consumption and time costs in the smelting process, and improving production efficiency. By controlling the viscosity of slag, various chemical reactions and physical processes in the metallurgical process can be optimized, thereby improving the quality and purity of metal products. Therefore, controlling the appropriate slag viscosity can optimize the metallurgical process, improve production efficiency, extend equipment life and promote comprehensive resource utilization, which is of great significance for optimizing steel production processes and improving product quality.

[0003] Experimental research and model calculation are the two main methods for obtaining viscosity data. Although experimental research to obtain slag viscosity can provide test results that are closer to the real environment, its high cost, high energy consumption, high requirements, operational complexity and potential safety risks are all factors that need to be considered. With the deepening of the understanding of slag systems, the use of models to predict viscosity has received more and more attention. There are two types of viscosity prediction models: empirical models and structural models. The empirical model is to regress and fit the slag composition and viscosity. In order to obtain higher accuracy, a large number of model parameters are introduced, and there is a lack of theoretical basis. The structural model is to establish a connection between the viscosity of the slag and the microstructure, but due to different understandings of the structure, the structural model also manifests itself in various forms. Model calculation can effectively improve R&D efficiency, reduce costs, and provide valuable reference for actual production. At present, a large number of scholars have established models for predicting viscosity. Common viscosity models include the classic Urbain model, Weymann-Frenkel model, Iida model, Nakamoto model, etc. The Urbain model is a model based on liquid dynamics theory. This model has a good predictive effect on the conventional CaO-SiO2-Al2O3-MgO quaternary system, but the prediction effect on other systems is average. The Weymann-Frenkel model can predict the melt viscosity in a wide range, taking into account the flow characteristics of the melt, but not its structural characteristics. The Iida model associates the basicity index of the melt with its structure. The applicable range of slag composition and temperature of this model is small. The Nakamoto model can predict the viscosity of the ternary slag system in a wide range of components, but there is a slight deviation in the prediction effect. In short, although predecessors have studied the viscosity and melt structure of slag and established corresponding viscosity models, due to the different systems on which the models are based, the application range of the models is narrow, the temperature range is small, and the prediction effect is poor.

[0004] Chemical composition is one of the main factors affecting slag viscosity. From a microscopic perspective, viscosity depends on the size of the ion clusters inside the melt and the strength of the interaction between them. Metallurgical slag is mostly silicate melt. According to the existence form of ions in the slag, the elements in the slag can be divided into three types: network formers, network modifiers and amphoterics. There are three types of oxygen: bridging oxygen (BO), non-bridging oxygen (NBO) and free oxygen (FO); oxygen that connects two forming bodies at the same time (TO b -T) is a bridging oxygen, an oxygen connected to a forming body and a modifying body (TO nb -M) is a non-bridging oxygen, and free oxygen is the oxygen ion with both ends connected to the modified oxygen (MO b -M). The slag melt is a network structure composed of [TO4]-tetrahedrons combined with network formers and oxygen, which contains different bridging oxygens, non-bridging oxygens and free oxygen. Tetrahedron is the basic structural unit in slag, usually represented by Qn Indicates that n = 0 to 4 represents the degree of oxygen bridging. The larger n is, the higher the Q T n The more complex it is, the more complex the system structure is. The larger n is, the more difficult it is for the unit to flow in the slag, and the higher the viscosity is; that is, the change in microstructure also represents the change in viscosity. By clarifying the microstructure of the slag, the viscosity of the slag can be predicted.

[0005] The microstructure information of slag is mainly obtained through spectral analysis and computer simulation methods. Spectral analysis directly obtains the microstructure information of slag through coordination, vibration, rotation and other information. Simulation mainly obtains the microstructure information in slag through molecular dynamics simulation. The rapid progress of science and technology and the continuous improvement of computational materials theory have led to the rapid development of molecular dynamics simulation and its wide application in the field of metallurgy. With the advent of the big data era, artificial intelligence based on a large amount of data has developed rapidly. Its application in the field of metallurgy can well promote the automation and intelligent development of the steel industry. Machine learning is a branch of artificial intelligence. This technology can explore the influencing factors of material properties, capture the changing laws of material properties, and effectively and accurately determine various process parameters. Machine learning technology can also analyze massive amounts of data information, efficiently and accurately handle complex nonlinear problems, and achieve fast and accurate prediction of target data. The research directions of machine learning mainly include decision trees, random forests, artificial neural networks, Bayesian learning and other aspects. The decision tree algorithm is famous for its use of the characteristics of tree structure. The path from the root node to the leaf node corresponds to a classification rule. Each leaf node represents a certain category. A significant advantage of decision trees is their intuitive structure and fast data processing speed. For example, XGBoost is an ensemble learning algorithm based on decision trees that combines efficiency, flexibility, and regularization mechanisms to prevent overfitting, allowing it to accurately deal with regression problems. When establishing a viscosity prediction model, the two main input features currently considered are composition and temperature. In fact, the essence of the effect of composition on viscosity lies in the change of microstructure. Therefore, if this microstructural information can be added to the model as an additional feature, a more accurate viscosity prediction model with wider application potential will be obtained. In other words, this helps to improve the directionality and accuracy of the prediction model, thereby broadening its scope of application in practical applications.

[0006] In summary, model calculation is an important way to obtain viscosity data. However, the existing viscosity models often have limitations due to the different systems on which they are based, such as narrow application areas, limited applicable temperature ranges, and poor prediction results. It is worth noting that viscosity is closely related to microstructure, and changes in microstructure directly reflect changes in viscosity. Therefore, obtaining the microstructural information of slag through spectral analysis and computer simulation methods, and combining machine learning technology to construct a viscosity prediction model containing microstructural information is expected to greatly improve the accuracy and scope of application of the model, which has important application value and practical significance for promoting the intelligent and efficient production of the metallurgical industry. Summary of the invention

[0007] (I) Purpose of the invention

[0008] In view of the problems existing in the above-mentioned technology, the purpose of the present invention is to provide a method for predicting the viscosity of metallurgical slag using machine learning, which adopts a method combining experimental detection and molecular dynamics simulation to obtain the microstructure information of metallurgical slag, adds the microstructure as a characteristic variable, and establishes a viscosity prediction model with slag composition, temperature and microstructure information as input variables. The application scope of the viscosity model is expanded, and it is pointed out that the viscosity of metallurgical slag under different composition conditions can be predicted clearly, quickly and accurately, which significantly improves the efficiency and accuracy of metallurgical slag viscosity prediction.

[0009] (II) Technical solution

[0010] In order to achieve the above object, the present invention provides a method for predicting the viscosity of metallurgical slag using molecular dynamics simulation, comprising the following steps:

[0011] S1. First, the experimental test data of various scholars on the microstructure of metallurgical slag were extracted and collected from the literature;

[0012] S2. Secondly, according to the specific components of the collected microstructure detection data, the molecular dynamics simulation method is used to supplement the parts with low data density;

[0013] S3. Finally, a machine learning method is used to establish a viscosity prediction model for the slag system with slag composition, temperature and microstructure as characteristic variables to obtain the predicted value of the slag viscosity to be tested.

[0014] Furthermore, in the step S1, the microstructure information of the metallurgical slag to be tested having the components of SiO2, Al2O3, CaO and MgO is obtained by searching relevant research on the microstructure of metallurgical slag published in academic journals.

[0015] Furthermore, in the step S1, the collected microstructure information includes the types of structural units and their molar contents of the slag system at different compositions and temperatures.

[0016] Furthermore, in step S2, based on the collected data component points, Lammps software is used to perform molecular dynamics simulation on the part with low data density to supplement the data.

[0017] Furthermore, in the step S2, the in file of the Lammps software running command executed also includes boundary conditions, equilibrium ensemble, temperature control method, pressure control method, cooling rate, summation method and relaxation time.

[0018] Furthermore, the boundary conditions adopt periodic boundary conditions, the ensemble is an NVT canonical ensemble, the temperature control method is a Nose-Hoover heat bath method, the pressure control method is a Parrinello-Rahman method, and the summation method is an Ewald summation method.

[0019] Furthermore, in step 2, the equilibrium structure data is read using statistical analysis software Matlab, the atomic information corresponding to each time step is recorded, and each type of atom is distinguished, and the distance between each oxygen atom, silicon atom and aluminum atom in the slag is calculated.

[0020] Furthermore, in step 2, the oxygen atom is taken as the center, and the cutoff distance of the Si-O bond and the Al-O bond is taken as the radius, and the silicon atoms and aluminum atoms around a single oxygen atom are counted. Then, the number of bridging oxygen atoms around each silicon atom (aluminum atom) is determined to obtain the structural unit Q Si n and Q Al n The molar content of .

[0021] Furthermore, in step S3, the model is constructed using the Visual Studio Code platform in the Python language environment and the Jupyter lab development environment.

[0022] Furthermore, in step S3, the collected data set is processed, outliers in the data set are processed using a deletion method, and missing values ​​are filled with a mode or left unprocessed.

[0023] Furthermore, in step S3, the data is selected for z-score standardization, and the StandardScaler function of the preprocessing module in the sklearn framework is called to perform z-score standardization on the data. The calculation formula is as follows:

[0024]

[0025] In the formula, x0 is the standardized feature variable, x is the original feature variable, x is the mean value of the feature variable, and Z is the standard deviation of the feature variable.

[0026] Furthermore, in step S3, the process of training the model requires feature selection, that is, selecting the optimal subset of variables that have an impact on the target variable or have predictive power from all variables in the data set. Use the Pearson correlation coefficient for feature selection:

[0027]

[0028] In the formula, r is the characteristic correlation coefficient, which indicates the linear correlation between the characteristic variable and viscosity, n is the number of samples, and x is i is the true value of each feature, is the average value of each feature, y i is the true value of viscosity, is the average viscosity.

[0029] Furthermore, in step S3, the molar content, temperature and microstructure unit Q of the slag components SiO2, CaO, MgO and Al2O3 are selected. n The mole fraction of is taken as the characteristic variable.

[0030] Furthermore, in step S3, data processing libraries such as Numpy and Pandas are used to process data, and the drop function is used to divide the slag components SiO2, CaO, MgO, temperature and the molar content of Al2O3 and the different structural units Q in the slag. n The mole fraction of is taken as the input variable and the slag viscosity is taken as the output variable.

[0031] Furthermore, in step S3, an intelligent algorithm XGBoost (eXtreme Gradient Boosting) based on a supervised classification learning method of Gradient Boosting is used to establish a microstructure model of metallurgical slag. The XGBoost algorithm supports custom loss functions and evaluation criteria, enabling it to cope with a variety of prediction problems. It is an additive model that only optimizes the sub-model in the current step in each iteration, such as in step n:

[0032] F n (x i )=F n-1 (x i )+f n (x i ) (1.4)

[0033] f n (x i ) is the sub-model of the current step, F n-1 (xi ) are the first n-1 sub-models that have been fixed after training.

[0034] Furthermore, in step S3, the train_test_split function of the preprocessing block is used to divide the model into a training set and a test set, accounting for 80% and 20% respectively. The training set is used for model training, and the test set is used to evaluate the accuracy of the trained model, and then the trained slag viscosity intelligent prediction model is obtained to obtain the predicted value of the slag viscosity to be tested.

[0035] (III) Beneficial effects

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention collects the melt microstructure of slag with different contents, fills in the missing data through molecular dynamics simulation, obtains the microstructure information, and then uses machine learning to calculate the molar content of slag components SiO2, CaO, MgO and Al2O3, slag temperature and different microstructure units Q in the slag. n The mole fraction of is taken as the input variable, and the slag viscosity is taken as the output variable, a slag viscosity model based on the microstructure is established, and the viscosity data of the slag to be measured is obtained. Therefore, the present invention proposes for the first time a method for intelligently predicting the viscosity of metallurgical slag based on the microstructure. The viscosity depends on the changes in the microstructure of the slag. The viscosity prediction model is established by taking the slag microstructure unit information calculated by molecular dynamics simulation as the characteristic variable, so as to achieve multi-characteristic variable, clear and accurate prediction of the slag viscosity, which can provide a reference for designing slag with reasonable smelting performance and promote the efficiency and intelligence of the metallurgical industry.

[0038] The present invention proposes a viscosity prediction method for metallurgical slag, and establishes an intelligent prediction model for slag viscosity based on microstructure. The constructed viscosity prediction model can accurately and efficiently calculate the viscosity value under specific conditions, which not only saves tedious and complicated experimental steps, but also saves experimental materials and time. Moreover, this method can predict the viscosity of slag under different composition conditions with high accuracy, and has great practicality and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention is described with the aid of the following drawings:

[0040] Figure 1 This is a flow chart of a method for predicting metallurgical slag by machine learning according to Example 1 of the present invention;

[0041] Figure 2 The figure is a relationship curve diagram between the predicted viscosity value and the measured viscosity value according to Example 1 of the present invention. DETAILED DESCRIPTION

[0042] In order to better understand the technical solution of the present invention, the content of the present invention includes but is not limited to the specific implementation methods described below, and similar technologies and methods should be considered to be within the scope of protection of the present invention. In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0043] It should be clear that the embodiments described in the present invention are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "an" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0045] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0046] Example 1

[0047] The present invention collects 358 sets of metallurgical slag microstructure information published in high-level academic journals at home and abroad, designs 20 sets of slag CaO-SiO2-Al2O3-MgO systems in areas with low data density, and uses molecular dynamics simulation to obtain their microstructure information. The specific components are shown in Table 1:

[0048] Table 1 Composition of experimental slag of CaO-SiO2-Al2O3-MgO system (mass fraction, %)

[0049]

[0050]

[0051] The molar ratio is obtained based on the mass ratio of each component of the slag (CaO, SiO2, Al2O3 and MgO), and then the number of different atoms (Si, Al, Ca, Mg and O) is determined based on the total number of atoms of 8000. Then the density of the slag is obtained based on experimental tests and the side length of the box is calculated, as shown in Table 2.

[0052] Table 2 Atomic number, density and box side length of experimental slag of CaO-SiO2-Al2O3-MgO system

[0053]

[0054] (1) Based on the above atomic types, numbers and slag density, a randomly distributed amorphous structure model is constructed. The initial structure data file and potential function of the slag to be tested are imported into the in file that executes the Lammps software running command. The initial structure of the slag to be tested is energy minimized in the Lammps software, and periodic boundary conditions are used to form a periodic image. The shape of the simulated cubic box remains unchanged, and the NVT ensemble is selected to maintain its stability. The Parrinello-Rahman and Nose-Hoover heat bath methods are used for pressure and temperature control. In terms of the algorithm, the cutoff radius of the short-range force is set to The cutoff diameter is smaller than the side length of any system box. The Ewald algorithm is used for long-range Coulomb force. The running time step is set to 1fs, and the data is saved every 10 steps through the frog leaping algorithm, for a total of 800,000 steps. In terms of temperature setting, the initial temperature is first set to 5000K, and 200,000 steps are run to fully mix the system to eliminate the initial distribution state of the atoms. Then, the temperature is reduced to 1773K through 200,000 steps, and finally, the temperature is continued to relax at 1773K for 400,000 steps, of which the first 20,000 steps are equilibrium relaxation. All data collection is completed within the last 200,000 steps, and finally the coordinate position file of 8,000 atoms in the system is output.

[0055] (2) Matlab software was used to read the equilibrium structure data, record the atomic information corresponding to each time step, distinguish the atoms of different types, calculate the distance between each oxygen atom, silicon atom and aluminum atom, and thus obtain the different structural units Q i The concentration distribution of (where i represents the number of bridging oxygen in the structural unit) is shown in Table 3.

[0056] Table 3 Structural unit Q of CaO-SiO2-Al2O3-MgO system experimental slag i content

[0057]

[0058] (3) The decision tree-based XGBoost algorithm was used to establish the microstructure model of metallurgical slag. In the Python language environment and the Jupyter lab development environment, the Visual Studio Code platform was used to process data based on the XGBoost framework using data processing libraries such as Numpy and Pandas. The data was standardized using the StandardScaler of the preprocessing module in the sklearn framework. The train_test_split function of the preprocessing block was used to divide the model training set and test set, accounting for 80% and 20% respectively. The training set was then used for model training, and the test set was used to evaluate the accuracy of the trained model, thereby obtaining a microstructure prediction model for the trained slag system under different components.

[0059] (4) 30 sets of viscosity values ​​were extracted from the 358 sets of test data and 20 sets of molecular dynamics simulation data, and the model of the present invention was used for calculation. The coefficient of determination (R 2 The viscosity prediction model was evaluated by three prediction evaluation indicators: mean square error (MSE) and root mean square error (RMSE), which were 0.8538, 0.0315 and 0.0010 respectively.

[0060] (5) The comparison between the viscosity values ​​calculated using this model and the experimental values ​​is shown in Figure 2 , the error is within 15%. At the same time, the Urbain model and Nakamoto model widely used in the metallurgical industry were also used for comparison. The results showed that the prediction of the model of the present invention was 3.8% higher than that of the Urbain model, and 2.4% higher than that of the Nakamoto model, which verified the accuracy of the model of the present invention.

[0061] Based on the above analysis, the viscosity prediction method of the present invention is applicable to all metallurgical slags with SiO2, Al2O3, CaO and MgO as main components. The present invention fully considers the influence mechanism of slag microstructure changes on fluidity, combines the microstructure information of the slag, adopts machine learning intelligent algorithm, and establishes a viscosity prediction model that depends on microstructure changes.

[0062] The above description shows and describes the preferred embodiments of the present invention, but as mentioned above, it should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the application concept described herein through the above teachings or the technology or knowledge of the relevant field. And the changes and modifications made by those skilled in the art do not depart from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for intelligently predicting metallurgical slag viscosity based on microstructure, characterized in that: The steps include: S1. First, the experimental test data of various scholars on the microstructure of metallurgical slag were extracted and collected from the literature; S2. Secondly, according to the specific components of the collected microstructure detection data, the molecular dynamics simulation method is used to supplement the parts with low data density; S3. Finally, a machine learning method is used to establish a viscosity prediction model for the slag system with slag composition, temperature and microstructure as characteristic variables to obtain the predicted value of the slag viscosity to be tested.

2. The method according to claim 1, characterized in that: In the step S1, the microstructure information of the metallurgical slag to be tested whose components are SiO2, Al2O3, CaO and MgO is obtained by searching the relevant research on the microstructure of metallurgical slag published in academic journals.

3. The method according to claim 1, characterized in that In the step S1, the collected microstructure information includes the types of structural units and their molar contents of the slag system at different compositions and temperatures.

4. The method according to claim 1, characterized in that In the step S2, the data is supplemented by using the method of molecular dynamics simulation of the part with low data density using Lammps software according to the component points of the collected data.

5. The method according to claim 1, characterized in that In the step S2, the in file of the Lammps software running command also includes boundary conditions, equilibrium ensemble, temperature control method, pressure control method, cooling rate, summation method and relaxation time.

6. The method according to claim 5, characterized in that The boundary conditions adopt periodic boundary conditions, the ensemble is the NVT canonical ensemble, the temperature control method is the Nose-Hoover heat bath method, the pressure control method is the Parrinello-Rahman method, and the summation method is the Ewald summation method.

7. The method according to claim 1, characterized in that In step S2, the equilibrium structure data is read using the statistical analysis software Matlab, the atomic information corresponding to each time step is recorded, and each type of atom is distinguished, and the distance between each oxygen atom, silicon atom and aluminum atom in the slag is calculated.

8. The method according to claim 7, characterized in that With the oxygen atom as the center and the cutoff distance of the Si-O bond and Al-O bond as the radius, the silicon atoms and aluminum atoms around a single oxygen atom are counted. Then the structural unit Q is determined by determining the number of bridging oxygens around each silicon atom (aluminum atom). n type and mole fraction.

9. The method according to claim 1, characterized in that: In step S3, the model is constructed using the Visual Studio Code platform in the Python language environment and the Jupyter lab development environment.

10. The method according to claim 1, characterized in that In step S3, the collected data set is processed, outliers in the data set are processed using a deletion method, and missing values ​​are filled with a mode or left unprocessed.

11. The method according to claim 1, characterized in that: In step S3, the data is selected for z-score standardization, and the StandardScaler function of the preprocessing module in the sklearn framework is called to perform z-score standardization on the data. The calculation formula is as follows: In the formula, x0 is the standardized feature variable, x is the original feature variable, is the mean value of the feature variable, and Z is the standard deviation of the feature variable.

12. The method according to claim 1, characterized in that In step S3, the process of training the model requires feature selection, that is, selecting the optimal variable subset that has an impact on the target variable or has predictive power from all variables in the data set. Use the Pearson correlation coefficient for feature selection: In the formula, r is the characteristic correlation coefficient, which indicates the linear correlation between the characteristic variable and viscosity, n is the number of samples, and x is i is the true value of each feature, is the average value of each feature, y i is the true value of viscosity, is the average viscosity.

13. The method according to claim 1, characterized in that In step S3, the molar contents of slag components SiO2, CaO, MgO and Al2O3, the microstructure unit Q n The mole fraction of and slag temperature are taken as characteristic variables.

14. The method according to claim 1, characterized in that In step S3, data processing libraries such as Numpy and Pandas are used to process data, and the molar content of slag components SiO2, CaO, MgO and Al2O3, and the different structural units Q in the slag are divided using the drop function. n The mole fraction of and slag temperature are taken as input variables, and the slag viscosity is taken as the output variable.

15. The method according to claim 1, characterized in that In step S3, an intelligent algorithm XGBoost (eXtreme Gradient Boosting) based on the supervised classification learning method of GradientBoosting is used to establish a microstructure model of metallurgical slag. The XGBoost algorithm supports custom loss functions and evaluation criteria, enabling it to cope with a variety of prediction problems. It is an additive model that only optimizes the sub-model in the current step in each iteration, such as in step n: F n (x i )=F n-1 (x i )+f n (x i ) (1.3) Among them, f n (x i ) is the sub-model of the current step, F n-1 (x i ) are the first n-1 sub-models that have been fixed after training.

16. The method according to claim 1, characterized in that In step S3, the train_test_split function of the preprocessing block is used to divide the model into a training set and a test set, accounting for 80% and 20% respectively. The training set is used for model training, and the test set is used to evaluate the accuracy of the trained model, and then the trained slag viscosity intelligent prediction model is obtained to obtain the predicted value of the slag viscosity to be tested.

Images